[PATCH v2] Documentation: gpu: Mention the requirements for new properties

Thu May 27 15:53:26 UTC 2021

Hi Maxime,

On Tue, May 25, 2021 at 12:39:40PM +0200, Maxime Ripard wrote:
> On Mon, May 24, 2021 at 07:04:43AM +0300, Laurent Pinchart wrote:
> > On Thu, May 20, 2021 at 04:24:35PM +0200, Maxime Ripard wrote:
> > > New KMS properties come with a bunch of requirements to avoid each
> > > driver from running their own, inconsistent, set of properties,
> > > eventually leading to issues like property conflicts, inconsistencies
> > > between drivers and semantics, etc.
> > > 
> > > Let's document what we expect.
> > > 
> > > Cc: Alexandre Belloni <alexandre.belloni at bootlin.com>
> > > Cc: Alexandre Torgue <alexandre.torgue at foss.st.com>
> > > Cc: Alex Deucher <alexander.deucher at amd.com>
> > > Cc: Alison Wang <alison.wang at nxp.com>
> > > Cc: Alyssa Rosenzweig <alyssa.rosenzweig at collabora.com>
> > > Cc: Andrew Jeffery <andrew at aj.id.au>
> > > Cc: Andrzej Hajda <a.hajda at samsung.com>
> > > Cc: Anitha Chrisanthus <anitha.chrisanthus at intel.com>
> > > Cc: Benjamin Gaignard <benjamin.gaignard at linaro.org>
> > > Cc: Ben Skeggs <bskeggs at redhat.com>
> > > Cc: Boris Brezillon <bbrezillon at kernel.org>
> > > Cc: Brian Starkey <brian.starkey at arm.com>
> > > Cc: Chen Feng <puck.chen at hisilicon.com>
> > > Cc: Chen-Yu Tsai <wens at csie.org>
> > > Cc: Christian Gmeiner <christian.gmeiner at gmail.com>
> > > Cc: "Christian König" <christian.koenig at amd.com>
> > > Cc: Chun-Kuang Hu <chunkuang.hu at kernel.org>
> > > Cc: Edmund Dea <edmund.j.dea at intel.com>
> > > Cc: Eric Anholt <eric at anholt.net>
> > > Cc: Fabio Estevam <festevam at gmail.com>
> > > Cc: Gerd Hoffmann <kraxel at redhat.com>
> > > Cc: Haneen Mohammed <hamohammed.sa at gmail.com>
> > > Cc: Hans de Goede <hdegoede at redhat.com>
> > > Cc: "Heiko Stübner" <heiko at sntech.de>
> > > Cc: Huang Rui <ray.huang at amd.com>
> > > Cc: Hyun Kwon <hyun.kwon at xilinx.com>
> > > Cc: Inki Dae <inki.dae at samsung.com>
> > > Cc: Jani Nikula <jani.nikula at linux.intel.com>
> > > Cc: Jernej Skrabec <jernej.skrabec at siol.net>
> > > Cc: Jerome Brunet <jbrunet at baylibre.com>
> > > Cc: Joel Stanley <joel at jms.id.au>
> > > Cc: John Stultz <john.stultz at linaro.org>
> > > Cc: Jonas Karlman <jonas at kwiboo.se>
> > > Cc: Jonathan Hunter <jonathanh at nvidia.com>
> > > Cc: Joonas Lahtinen <joonas.lahtinen at linux.intel.com>
> > > Cc: Joonyoung Shim <jy0922.shim at samsung.com>
> > > Cc: Jyri Sarha <jyri.sarha at iki.fi>
> > > Cc: Kevin Hilman <khilman at baylibre.com>
> > > Cc: Kieran Bingham <kieran.bingham+renesas at ideasonboard.com>
> > > Cc: Krzysztof Kozlowski <krzysztof.kozlowski at canonical.com>
> > > Cc: Kyungmin Park <kyungmin.park at samsung.com>
> > > Cc: Laurent Pinchart <Laurent.pinchart at ideasonboard.com>
> > > Cc: Linus Walleij <linus.walleij at linaro.org>
> > > Cc: Liviu Dudau <liviu.dudau at arm.com>
> > > Cc: Lucas Stach <l.stach at pengutronix.de>
> > > Cc: Ludovic Desroches <ludovic.desroches at microchip.com>
> > > Cc: Marek Vasut <marex at denx.de>
> > > Cc: Martin Blumenstingl <martin.blumenstingl at googlemail.com>
> > > Cc: Matthias Brugger <matthias.bgg at gmail.com>
> > > Cc: Maxime Coquelin <mcoquelin.stm32 at gmail.com>
> > > Cc: Maxime Ripard <mripard at kernel.org>
> > > Cc: Melissa Wen <melissa.srw at gmail.com>
> > > Cc: Neil Armstrong <narmstrong at baylibre.com>
> > > Cc: Nicolas Ferre <nicolas.ferre at microchip.com>
> > > Cc: "Noralf Trønnes" <noralf at tronnes.org>
> > > Cc: NXP Linux Team <linux-imx at nxp.com>
> > > Cc: Oleksandr Andrushchenko <oleksandr_andrushchenko at epam.com>
> > > Cc: Patrik Jakobsson <patrik.r.jakobsson at gmail.com>
> > > Cc: Paul Cercueil <paul at crapouillou.net>
> > > Cc: Pengutronix Kernel Team <kernel at pengutronix.de>
> > > Cc: Philippe Cornu <philippe.cornu at foss.st.com>
> > > Cc: Philipp Zabel <p.zabel at pengutronix.de>
> > > Cc: Qiang Yu <yuq825 at gmail.com>
> > > Cc: Rob Clark <robdclark at gmail.com>
> > > Cc: Robert Foss <robert.foss at linaro.org>
> > > Cc: Rob Herring <robh at kernel.org>
> > > Cc: Rodrigo Siqueira <rodrigosiqueiramelo at gmail.com>
> > > Cc: Rodrigo Vivi <rodrigo.vivi at intel.com>
> > > Cc: Roland Scheidegger <sroland at vmware.com>
> > > Cc: Russell King <linux at armlinux.org.uk>
> > > Cc: Sam Ravnborg <sam at ravnborg.org>
> > > Cc: Sandy Huang <hjc at rock-chips.com>
> > > Cc: Sascha Hauer <s.hauer at pengutronix.de>
> > > Cc: Sean Paul <sean at poorly.run>
> > > Cc: Seung-Woo Kim <sw0312.kim at samsung.com>
> > > Cc: Shawn Guo <shawnguo at kernel.org>
> > > Cc: Stefan Agner <stefan at agner.ch>
> > > Cc: Steven Price <steven.price at arm.com>
> > > Cc: Sumit Semwal <sumit.semwal at linaro.org>
> > > Cc: Thierry Reding <thierry.reding at gmail.com>
> > > Cc: Tian Tao <tiantao6 at hisilicon.com>
> > > Cc: Tomeu Vizoso <tomeu.vizoso at collabora.com>
> > > Cc: Tomi Valkeinen <tomba at kernel.org>
> > > Cc: VMware Graphics <linux-graphics-maintainer at vmware.com>
> > > Cc: Xinliang Liu <xinliang.liu at linaro.org>
> > > Cc: Xinwei Kong <kong.kongxinwei at hisilicon.com>
> > > Cc: Yannick Fertre <yannick.fertre at foss.st.com>
> > > Cc: Zack Rusin <zackr at vmware.com>
> > > Reviewed-by: Daniel Vetter <daniel.vetter at ffwll.ch>
> > > Signed-off-by: Maxime Ripard <maxime at cerno.tech>
> > > 
> > > ---
> > > 
> > > Changes from v2:
> > >   - Typos and wording reported by Daniel and Alex
> > > ---
> > >  Documentation/gpu/drm-kms.rst | 19 +++++++++++++++++++
> > >  1 file changed, 19 insertions(+)
> > > 
> > > diff --git a/Documentation/gpu/drm-kms.rst b/Documentation/gpu/drm-kms.rst
> > > index 87e5023e3f55..c28b464dd397 100644
> > > --- a/Documentation/gpu/drm-kms.rst
> > > +++ b/Documentation/gpu/drm-kms.rst
> > > @@ -463,6 +463,25 @@ KMS Properties
> > >  This section of the documentation is primarily aimed at user-space developers.
> > >  For the driver APIs, see the other sections.
> > >  
> > > +Requirements
> > > +------------
> > > +
> > > +KMS drivers might need to add extra properties to support new features.
> > > +Each new property introduced in a driver need to meet a few
> > 
> > s/need/needs/
> > 
> > > +requirements, in addition to the one mentioned above.:
> > 
> > s/above./above/
> > 
> > > +
> > > +- It must be standardized, with some documentation to describe how the
> > > +  property can be used.
> > > +
> > > +- It must provide a generic helper in the core code to register that
> > > +  property on the object it attaches to.
> > > +
> > > +- Its content must be decoded by the core and provided in the object's
> > > +  associated state structure. That includes anything drivers might want to
> > > +  precompute, like :c:type:`struct drm_clip_rect <drm_clip_rect>` for planes.
> > 
> > Does this effectively mean that we completely forbid driver-specific
> > properties ? While I agree that we should strive for standardization,
> > there are two issues that worry me. The first one is simple, we may need
> > to control features that would be very device-specific, and
> > standardizing properties doesn't seem to make much sense in that case.
> 
> I'd say that we should make it clear in that case that it's
> driver-specific.
> 
> > The second issue relates to properties that could be applicable to
> > multiple devices, but for which we have a single driver. Designing a
> > standard with a single data point usually leads to a bad design. I'm not
> > sure how to handle this correctly though, as we certainly don't want
> > this to be taken as an excuse to create driver-specific properties when
> > generic properties would make sense.
> 
> The discussion that made us create that patch was about this property:
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/sti/sti_hdmi.c#n170
> 
> It's all kind of bad:
>   - It kind of conflicts with the generic Colorspace property
>   - It's not really a colorspace (Not that "Colorspace" is either)
>   - It could have been made generic from the start
>   - We don't have any knowledge on who uses it and why, so it's
>     difficult to rework
> 
> This was introduced before we had any kind of rule or documentation on
> the UAPI though, so there's no-one to blame really but we don't really
> want to have something like that happen again.
> 
> I agree that doing something generic from the beginning can be
> difficult, but this is some userspace API that we will have to carry
> around forever, so it's worth it I guess?

It is, no disagreement about that. Pushing driver authors to explore
standardization of properties is a good idea. As long as we have a
pragmatic approach and allow vendor-specific properties when it makes
sense, I'll have no concern.

> You have a point on the vendor properties though. Maybe we can require a
> vendor prefix for those? It would reduce the risk of a conflict.

I like the idea of a vendor prefix.

-- 
Regards,

Laurent Pinchart